Brand Voice: development of brand communication tone

Voice and tone are not the same, and confusing them is costly

Voice is the brand’s personality, fixed for the long term: if a company speaks with humor and without jargon, it won’t suddenly start sounding like a government agency in a quarter. Tone is the mood within this personality, which changes with the situation. A person with a stable character also has different moods at a funeral and at a birthday — the character is the same, the tone is different.

Confusion arises when a team perceives voice as a rigid template of phrases and tries to force it onto any context without changes. The result is a service failure message written in the same cheerful tone as a promo for a new feature. This is not voice, this is deafness to the situation.

What voice consists of in practice

Brand voice is not only copywriting for advertising banners. It penetrates where text is formally scarce, but it is precisely there that it is decided whether the user trusts the company further.

• Copywriting — website texts, articles, advertising campaigns, where the voice sounds most developed.
• Microtexts — buttons, form errors, notifications, interface labels. It is here that the voice either strengthens or collapses: 'Oops, something went wrong' and 'Error 404: page not found' — two different brands.
• Social networks — the most live channel where the voice is tested on real information flows in real time.
• Customer support — correspondence with support is often more honest than advertising: there's either a voice or it's a script.
• Visual signatures and delivery — font, response speed, emojis or their absence — all are part of the voice, even though formally they are not text.

Unisender conveys care through supportive and hinting formulations in the interface. Sber maintains a neutral-business tone even in chatbots — this is a conscious decision for an audience that values bank predictability over wit.

How to adapt the voice for channels without losing it along the way

The problem is usually one: the team decides that LinkedIn needs a 'serious' brand, and Instagram needs a 'funny' one, and in the end, they get two different brands instead of one voice in different clothes. The right approach is to fix not a set of phrases, but a system of allowable deviations.

1. Define voice as the core — 3-5 characteristics that never change.
2. For each channel, describe the allowable tone range: how formal or relaxed one can be without stepping outside the brand's identity.
3. Specify the response to typical situations — complaint, failure, celebration, crisis — separately for each channel.
4. Check the extreme points: the most informal brand message and the most formal — they should be recognizable as one voice.

SOOQA.store maintains a panibratistic, harsh tone in social networks and email newsletters — the difference is only in the intensity of sarcasm, not in its presence. This is managed adaptation, not two different brands under one sign.

Voice cannot be invented in a vacuum — it must be tested

Most guides stop at the 'formulate characteristics' stage and move on to examples. That's half the job. Nielsen Norman Group found that variations in tone across four dimensions — formality, seriousness, respectfulness, enthusiasm — produce statistically significant differences in user perception, although the effect is modest: 0,5–1 points on a 5-point scale. Cold, dry, formal formulations are almost always read as less friendly — and this is not an opinion, but a measurable result.

Practical voice validation looks like this: 2-3 variants of the same message with different intonations are taken, shown to the target audience without indicating the brand, and measured to see which variant evokes more trust and desire to continue the dialogue. For UX texts, the bipolar NN/g grid is less effective than Emotional Journey Map — a map of the user's emotional state at the moment of interaction. A person reading a payment error message is emotionally in a different place than a person browsing the brand's social media feed — and the voice must take this into account, rather than remaining equally 'friendly' everywhere.

Typical mistakes — and why the guideline itself doesn't save

• Voice as a list of adjectives without examples. ‘Friendly, expert, modern’ — three words that don’t help a copywriter write a specific phrase.
• Lack of anti-examples. The formulation ‘we are X, but not Y’ works better than simply ‘we are X’, because it clearly excludes semantically similar but incorrect intonations.
• Voice without an owner. If no one in the company is responsible for ensuring the voice aligns with the brand, the document becomes a PDF that no one opens.
• Development without team training. Research shows: at least 60% of the success of voice implementation depends not on the quality of the document itself, but on how it was communicated to the team and maintained in practice. Without a control system, the guideline loses relevance within three months.
• Copying someone else's voice. The Slack tone or the Starbucks tone cannot be transferred mechanically — it grew from a specific audience and context that another company may not have.

What should really be in a voice guideline

A working document using the Asana approach is not a presentation with beautiful words, but a tool that copywriters refer to daily. It should contain:

• 3-5 characteristics of the voice with specific sound for each — not just the word 'friendly', but an example of a phrase that sounds friendly specifically for this brand;
• formulations of the type 'we are X, but not Y' — to exclude similar but incorrect intonations;
• examples of what not to do, with an explanation of why a specific phrase doesn't fit;
• examples of adaptation for different channels — how the brand sounds in support, social media, email;
• a simple formula for quick verification, like MailChimp's approach of 'funny but not clowny' — it helps the copywriter understand at a glance whether they've crossed the line.

A voice without metrics is a beautiful theory. Test it not only by the editor's feelings, but also by recognizability in blind tests, by engagement in social networks, and by feedback from support — it is there that the voice either works for the business or remains a slogan in presentations.